Kaiser window

From Wikipedia, the free encyclopedia

The Kaiser window is a window function wk used for digital signal processing, and is defined by the formula:

Kaiser window function for N=100 and α= 0.5,1,2,4,8,16.
Enlarge
Kaiser window function for N=100 and α= 0.5,1,2,4,8,16.
w_k = \left\{ \begin{matrix} \frac{I_0\left(\pi\alpha \sqrt{1 - (\frac{2k}{N}-1)^2}\right)} {I_0(\pi\alpha)} & \mbox{if } 0 \leq k \leq N \\ \\ 0 & \mbox{otherwise} \\ \end{matrix} \right.

where I0 is the zeroth order modified Bessel function of the first kind, α is an arbitrary real number that determines the shape of the window, and the integer N gives the length of the window (N+1 points).

By construction, this function peaks at unity for k = N/2, i.e. at the center of the window, and decays exponentially towards the window edges.

The larger the value of |α|, the narrower the window becomes; α = 0 corresponds to a rectangular window. Conversely, for larger |α| the width of the main lobe increases in the Fourier transform of wk, while the side lobes decrease in amplitude. Thus, this parameter controls the tradeoff between main-lobe width and side-lobe area, as is illustrated in the plot of the frequency spectra below. For large α, the shape of the Kaiser window (in both time and frequency domain) tends to a Gaussian curve. The Kaiser window is nearly optimal in the sense of its peak's concentration around ω=0 (Oppenheim et al., 1999).

Frequency spectra of Kaiser windows for α=2 and α=4. The sharp minima in the side lobes are places where the amplitude goes all the way to zero, but does not here because of the finite plotting resolution.
Enlarge
Frequency spectra of Kaiser windows for α=2 and α=4. The sharp minima in the side lobes are places where the amplitude goes all the way to zero, but does not here because of the finite plotting resolution.

[edit] Kaiser-Bessel derived (KBD) window

A related window function is the Kaiser-Bessel derived (KBD) window, which is designed to be suitable for use with the modified discrete cosine transform (MDCT). The KBD window function dn is defined in terms of the Kaiser window wk by the formula:

KBD window function for N=100 and α= 0.5, 2, 8, 32.
Enlarge
KBD window function for N=100 and α= 0.5, 2, 8, 32.
d_k = \left\{ \begin{matrix} \sqrt{\frac{\sum_{j=0}^{k} w_j} {\sum_{j=0}^{N} w_j}} & \mbox{if } 0 \leq k < N \\ \\ \sqrt{\frac{\sum_{j=0}^{2N-1-k} w_j} {\sum_{j=0}^{N} w_j}} & \mbox{if } N \leq k < 2N \\ \\ 0 & \mbox{otherwise} \\ \end{matrix} \right.

This defines a window of length 2N, where by construction dk satisfies the Princen-Bradley condition for the MDCT (using the fact that wN−k = wk): dk2 + dk + N2 = 1 (interpreting k and k + N modulo 2N). The KBD window is also symmetric in the proper manner for the MDCT: dk = d2N−1−k.

[edit] References

  • Oppenheim, A. V.; Schafer, R. W.; and Buck J. R. (1999). Discrete-time signal processing. Upper Saddle River, N.J.: Prentice Hall. ISBN 0-13-754920-2.
  • Kaiser, J. F. (1966). Digital Filters. In Kuo, F. F. and Kaiser, J. F. (Eds.), System Analysis by Digital Computer, chap. 7. New York, Wiley.
  • Marina Bosi, Kaiser-Bessel Derived Window, Music 422 / EE 367C: Perceptual Audio Coding (Stanford University course page, 2005).
Retrieved from "http://en.wikipedia.org../../../k/a/i/Kaiser_window.html"
In other languages